[ 
https://issues.apache.org/jira/browse/SPARK-16693?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15833206#comment-15833206
 ] 

Felix Cheung commented on SPARK-16693:
--------------------------------------

I want to bring this up again and potentially to discuss in dev@, now that we 
are on Spark 2.2.

Not only having these are harder to work with, making it harder to add new 
parameter (eg. numPartitions) and so on, but more importantly, the wrapper/stub 
methods (e.g createDataFrame.default) shows up in auto-complete, tooltips, help 
doc and so on and create confusion.

Moreover, on a slightly orthogonal note, we should also consider removing all 
the internal RDD methods, or at least making them non-export S3 methods. Right 
now every time we are adding a new method having the same name of the existing, 
internal-only RDD method in R we would need to update the generic, and rename 
the existing method (by appending "RDD" to its name), and all its call sites, 
otherwise we would get a check-cran warning of lacking documentation. And by 
adding such method to the NAMESPACE file the existing RDD-only will get exposed 
as well (again showing up in auto-complete etc.), unless it is renamed. Since 
we are renaming the existing method to fix this, we are also breaking backward 
compatibility anyway (although the method was not public so strictly speaking 
there has not been any guarantee).

But I could scope this JIRA to only sqlContext methods and leave this 2nd issue 
to a different JIRA.


> Remove R deprecated methods
> ---------------------------
>
>                 Key: SPARK-16693
>                 URL: https://issues.apache.org/jira/browse/SPARK-16693
>             Project: Spark
>          Issue Type: Bug
>          Components: SparkR
>    Affects Versions: 2.0.0
>            Reporter: Felix Cheung
>
> For methods deprecated in Spark 2.0.0, we should remove them in 2.1.0



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to